1 Using the initial RAM disk (initrd)
4 Written 1996,2000 by Werner Almesberger <firstname.lastname@example.org> and
5 Hans Lermen <email@example.com>
8 initrd provides the capability to load a RAM disk by the boot loader.
9 This RAM disk can then be mounted as the root file system and programs
10 can be run from it. Afterwards, a new root file system can be mounted
11 from a different device. The previous root (from initrd) is then moved
12 to a directory and can be subsequently unmounted.
14 initrd is mainly designed to allow system startup to occur in two phases,
15 where the kernel comes up with a minimum set of compiled-in drivers, and
16 where additional modules are loaded from initrd.
18 This document gives a brief overview of the use of initrd. A more detailed
19 discussion of the boot process can be found in .
25 When using initrd, the system typically boots as follows:
27 1) the boot loader loads the kernel and the initial RAM disk
28 2) the kernel converts initrd into a "normal" RAM disk and
29 frees the memory used by initrd
30 3) initrd is mounted read-write as root
31 4) /linuxrc is executed (this can be any valid executable, including
32 shell scripts; it is run with uid 0 and can do basically everything
33 init can do)
34 5) linuxrc mounts the "real" root file system
35 6) linuxrc places the root file system at the root directory using the
36 pivot_root system call
37 7) the usual boot sequence (e.g. invocation of /sbin/init) is performed
38 on the root file system
39 8) the initrd file system is removed
41 Note that changing the root directory does not involve unmounting it.
42 It is therefore possible to leave processes running on initrd during that
43 procedure. Also note that file systems mounted under initrd continue to
44 be accessible.
47 Boot command-line options
50 initrd adds the following new options:
52 initrd=<path> (e.g. LOADLIN)
54 Loads the specified file as the initial RAM disk. When using LILO, you
55 have to specify the RAM disk image file in /etc/lilo.conf, using the
56 INITRD configuration variable.
60 initrd data is preserved but it is not converted to a RAM disk and
61 the "normal" root file system is mounted. initrd data can be read
62 from /dev/initrd. Note that the data in initrd can have any structure
63 in this case and doesn't necessarily have to be a file system image.
64 This option is used mainly for debugging.
66 Note: /dev/initrd is read-only and it can only be used once. As soon
67 as the last process has closed it, all data is freed and /dev/initrd
68 can't be opened anymore.
70 root=/dev/ram0 (without devfs)
71 root=/dev/rd/0 (with devfs)
73 initrd is mounted as root, and the normal boot procedure is followed,
74 with the RAM disk still mounted as root.
80 First, a directory for the initrd file system has to be created on the
81 "normal" root file system, e.g.
83 # mkdir /initrd
85 The name is not relevant. More details can be found on the pivot_root(2)
86 man page.
88 If the root file system is created during the boot procedure (i.e. if
89 you're building an install floppy), the root file system creation
90 procedure should create the /initrd directory.
92 If initrd will not be mounted in some cases, its content is still
93 accessible if the following device has been created (note that this
94 does not work if using devfs):
96 # mknod /dev/initrd b 1 250
97 # chmod 400 /dev/initrd
99 Second, the kernel has to be compiled with RAM disk support and with
100 support for the initial RAM disk enabled. Also, at least all components
101 needed to execute programs from initrd (e.g. executable format and file
102 system) must be compiled into the kernel.
104 Third, you have to create the RAM disk image. This is done by creating a
105 file system on a block device, copying files to it as needed, and then
106 copying the content of the block device to the initrd file. With recent
107 kernels, at least three types of devices are suitable for that:
109 - a floppy disk (works everywhere but it's painfully slow)
110 - a RAM disk (fast, but allocates physical memory)
111 - a loopback device (the most elegant solution)
113 We'll describe the loopback device method:
115 1) make sure loopback block devices are configured into the kernel
116 2) create an empty file system of the appropriate size, e.g.
117 # dd if=/dev/zero of=initrd bs=300k count=1
118 # mke2fs -F -m0 -b 1024 initrd
119 (if space is critical, you may want to use the Minix FS instead of Ext2)
120 (Note that due to a problem elsewhere in the kernel, you _must_ use a
121 1024-byte blocksize when creating your file system. If any other
122 value is used, the kernel will be unable to mount the initrd at boot
123 time, causing a kernel panic.)
124 3) mount the file system, e.g.
125 # mount -t ext2 -o loop initrd /mnt
126 4) create the console device (not necessary if using devfs, but it can't
127 hurt to do it anyway):
128 # mkdir /mnt/dev
129 # mknod /mnt/dev/console c 5 1
130 5) copy all the files that are needed to properly use the initrd
131 environment. Don't forget the most important file, /linuxrc
132 Note that /linuxrc's permissions must include "x" (execute).
133 6) correct operation the initrd environment can frequently be tested
134 even without rebooting with the command
135 # chroot /mnt /linuxrc
136 This is of course limited to initrds that do not interfere with the
137 general system state (e.g. by reconfiguring network interfaces,
138 overwriting mounted devices, trying to start already running demons,
139 etc. Note however that it is usually possible to use pivot_root in
140 such a chroot'ed initrd environment.)
141 7) unmount the file system
142 # umount /mnt
143 8) the initrd is now in the file "initrd". Optionally, it can now be
145 # gzip -9 initrd
147 For experimenting with initrd, you may want to take a rescue floppy and
148 only add a symbolic link from /linuxrc to /bin/sh. Alternatively, you
149 can try the experimental newlib environment  to create a small
152 Finally, you have to boot the kernel and load initrd. Almost all Linux
153 boot loaders support initrd. Since the boot process is still compatible
154 with an older mechanism, the following boot command line parameters
155 have to be given:
157 root=/dev/ram0 init=/linuxrc rw
159 if not using devfs, or
161 root=/dev/rd/0 init=/linuxrc rw
163 if using devfs. (rw is only necessary if writing to the initrd file
166 With LOADLIN, you simply execute
168 LOADLIN <kernel> initrd=<disk_image>
169 e.g. LOADLIN C:\LINUX\BZIMAGE initrd=C:\LINUX\INITRD.GZ root=/dev/ram0
170 init=/linuxrc rw
172 With LILO, you add the option INITRD=<path> to either the global section
173 or to the section of the respective kernel in /etc/lilo.conf, and pass
174 the options using APPEND, e.g.
176 image = /bzImage
177 initrd = /boot/initrd.gz
178 append = "root=/dev/ram0 init=/linuxrc rw"
180 and run /sbin/lilo
182 For other boot loaders, please refer to the respective documentation.
184 Now you can boot and enjoy using initrd.
187 Changing the root device
190 When finished with its duties, linuxrc typically changes the root device
191 and proceeds with starting the Linux system on the "real" root device.
193 The procedure involves the following steps:
194 - mounting the new root file system
195 - turning it into the root file system
196 - removing all accesses to the old (initrd) root file system
197 - unmounting the initrd file system and de-allocating the RAM disk
199 Mounting the new root file system is easy: it just needs to be mounted on
200 a directory under the current root. Example:
202 # mkdir /new-root
203 # mount -o ro /dev/hda1 /new-root
205 The root change is accomplished with the pivot_root system call, which
206 is also available via the pivot_root utility (see pivot_root(8) man
207 page; pivot_root is distributed with util-linux version 2.10h or higher
208 ). pivot_root moves the current root to a directory under the new
209 root, and puts the new root at its place. The directory for the old root
210 must exist before calling pivot_root. Example:
212 # cd /new-root
213 # mkdir initrd
214 # pivot_root . initrd
216 Now, the linuxrc process may still access the old root via its
217 executable, shared libraries, standard input/output/error, and its
218 current root directory. All these references are dropped by the
219 following command:
221 # exec chroot . what-follows <dev/console >dev/console 2>&1
223 Where what-follows is a program under the new root, e.g. /sbin/init
224 If the new root file system will be used with devfs and has no valid
225 /dev directory, devfs must be mounted before invoking chroot in order to
226 provide /dev/console.
228 Note: implementation details of pivot_root may change with time. In order
229 to ensure compatibility, the following points should be observed:
231 - before calling pivot_root, the current directory of the invoking
232 process should point to the new root directory
233 - use . as the first argument, and the _relative_ path of the directory
234 for the old root as the second argument
235 - a chroot program must be available under the old and the new root
236 - chroot to the new root afterwards
237 - use relative paths for dev/console in the exec command
239 Now, the initrd can be unmounted and the memory allocated by the RAM
240 disk can be freed:
242 # umount /initrd
243 # blockdev --flushbufs /dev/ram0 # /dev/rd/0 if using devfs
245 It is also possible to use initrd with an NFS-mounted root, see the
246 pivot_root(8) man page for details.
248 Note: if linuxrc or any program exec'ed from it terminates for some
249 reason, the old change_root mechanism is invoked (see section "Obsolete
250 root change mechanism").
253 Usage scenarios
256 The main motivation for implementing initrd was to allow for modular
257 kernel configuration at system installation. The procedure would work
258 as follows:
260 1) system boots from floppy or other media with a minimal kernel
261 (e.g. support for RAM disks, initrd, a.out, and the Ext2 FS) and
262 loads initrd
263 2) /linuxrc determines what is needed to (1) mount the "real" root FS
264 (i.e. device type, device drivers, file system) and (2) the
265 distribution media (e.g. CD-ROM, network, tape, ...). This can be
266 done by asking the user, by auto-probing, or by using a hybrid
268 3) /linuxrc loads the necessary kernel modules
269 4) /linuxrc creates and populates the root file system (this doesn't
270 have to be a very usable system yet)
271 5) /linuxrc invokes pivot_root to change the root file system and
272 execs - via chroot - a program that continues the installation
273 6) the boot loader is installed
274 7) the boot loader is configured to load an initrd with the set of
275 modules that was used to bring up the system (e.g. /initrd can be
276 modified, then unmounted, and finally, the image is written from
277 /dev/ram0 or /dev/rd/0 to a file)
278 8) now the system is bootable and additional installation tasks can be
281 The key role of initrd here is to re-use the configuration data during
282 normal system operation without requiring the use of a bloated "generic"
283 kernel or re-compiling or re-linking the kernel.
285 A second scenario is for installations where Linux runs on systems with
286 different hardware configurations in a single administrative domain. In
287 such cases, it is desirable to generate only a small set of kernels
288 (ideally only one) and to keep the system-specific part of configuration
289 information as small as possible. In this case, a common initrd could be
290 generated with all the necessary modules. Then, only /linuxrc or a file
291 read by it would have to be different.
293 A third scenario are more convenient recovery disks, because information
294 like the location of the root FS partition doesn't have to be provided at
295 boot time, but the system loaded from initrd can invoke a user-friendly
296 dialog and it can also perform some sanity checks (or even some form of
299 Last not least, CD-ROM distributors may use it for better installation
300 from CD, e.g. by using a boot floppy and bootstrapping a bigger RAM disk
301 via initrd from CD; or by booting via a loader like LOADLIN or directly
302 from the CD-ROM, and loading the RAM disk from CD without need of
306 Obsolete root change mechanism
309 The following mechanism was used before the introduction of pivot_root.
310 Current kernels still support it, but you should _not_ rely on its
311 continued availability.
313 It works by mounting the "real" root device (i.e. the one set with rdev
314 in the kernel image or with root=... at the boot command line) as the
315 root file system when linuxrc exits. The initrd file system is then
316 unmounted, or, if it is still busy, moved to a directory /initrd, if
317 such a directory exists on the new root file system.
319 In order to use this mechanism, you do not have to specify the boot
320 command options root, init, or rw. (If specified, they will affect
321 the real root file system, not the initrd environment.)
323 If /proc is mounted, the "real" root device can be changed from within
324 linuxrc by writing the number of the new root FS device to the special
325 file /proc/sys/kernel/real-root-dev, e.g.
327 # echo 0x301 >/proc/sys/kernel/real-root-dev
329 Note that the mechanism is incompatible with NFS and similar file
332 This old, deprecated mechanism is commonly called "change_root", while
333 the new, supported mechanism is called "pivot_root".
339  Almesberger, Werner; "Booting Linux: The History and the Future"
341  newlib package (experimental), with initrd example
343  Brouwer, Andries; "util-linux: Miscellaneous utilities for Linux"
Cache object: 1347f0a22f86adc6c32a526ee88d9d52